Smart and authentic: How anthropomorphic features of AI chatbots drive reactance among Indian university students

This study integrates privacy calculus and psychological reactance theories to examine the mechanism by which the anthropomorphic characteristics of AI chatbots, such as perceived authenticity and competence, drive reactance through perceived privacy concerns and identity threat among Indian university students. Survey data from 449 students were analyzed using covariance-based structural equation modelling (CB-SEM). The results show that perceived authenticity is significantly associated with reactance in a complementary mediated relationship, alongside a direct positive association with reactance. Perceived competence operated through a competitive mediated relationship; while it indirectly increased reactance via perceived privacy concerns and identity threat, it also showed a significant, direct negative association with reactance. Findings suggest that students’ reactance to AI chatbots arises in response to anthropomorphic AI personas, triggered directly by the perceived invasiveness of authentic features and indirectly by the surveillance risks associated with competent AI chatbots. Finally, we argue for transparent, machine-like intelligence expansion, rather than deceptive human-like features, as key targets for future AI design and deployment.